Advances in molecular simulation methods and hardware have finally made possible the simulation of biomolecules on physiological timescales. Such simulations provide an important tool to aid the molecular understanding of disease. However, fully integrating these new tools into medical research requires that simulation software becomes easier to use, easier to develop for, and more hardened, reliable and robust. Making these capabilities accessible and reliable in the hands of a broad scientific community will require transformative gains in reliability, interoperability, flexibility, and ease of use atscale. We therefore propose improvements to the open-source Gromacs molecular simulation package to help meet these challenges. These improvements will 1) enable automated tests of physical validity for both code contributions and simulation parameter choices, 2) provide a simple interface for external code to utilize Gromacs functionality, and 3) provide a high-performance interface to mix Monte Carlo and molecular dynamics simulation. This work will facilitate community extension via a robust testing framework integrated with our code review system and via an API to allow implementation of new tools and simulations using high-level calls. These efforts will leverage best practices from software engineering and be implemented in a way to put tools in the hands of a broad developer and user community.